Long Term Evolution (LTE)
Third-generation mobile systems (3G) based on WCDMA (Wideband Code Division Multiple Access) radio-access technology are being deployed on a broad scale all around the world. A first step in enhancing or evolving this technology entails introducing High-Speed Downlink Packet Access (HSDPA) and an enhanced uplink, also referred to as High Speed Uplink Packet Access (HSUPA), giving a radio-access technology that is highly competitive.
However, knowing that user and operator requirements and expectations will continue to evolve, the 3GPP (3rd Generation Partnership Project) has begun considering the next major step or evolution of the 3G standard to ensure the long-term competitiveness of 3G. The 3GPP launched a Study Item “Evolved UTRA and UTRAN” (abbreviated E-UTRA and E-UTRAN) also referred to as long-term evolution (LTE). The study will investigate means of achieving major leaps in performance in order to improve service provisioning and reduce user and operator costs.
It is generally assumed that there will be a convergence toward the use of Internet Protocols (IP), and all future services will be carried on top of IP. Therefore, the focus of the evolution is on enhancements to the packet-switched (PS) domain.
The main objectives of the evolution are to further improve service provisioning and reduce user and operator costs as already mentioned.
More specifically, some key performance and capability targets for the long-term evolution are
Significantly higher data rates compared to HSDPA and HSUPA: envisioned target peak data rates of more than 100 Mbps over the downlink and 50 Mbps over the uplink
Improved coverage: high data rates with wide-area coverage
Significantly reduced latency in the user plane in the interest of improving the performance of higher layer protocols (for example, TCP) as well as reducing the delay associated with control plane procedures (for instance, session setup)
Greater system capacity: threefold capacity compared to current standards.
One other key requirement of the long-term evolution is to allow for a smooth migration to these technologies.
LTE Architecture
In FIG. 1 an overview of a 3GPP LTE mobile communication network is shown. The network consists of different network entities that are functionally grouped into the Evolved Packet Core (EPC), the Radio Access Network (RAN) and the User Equipments (UEs) or mobile terminals.
The radio access network is responsible for handling all radio-related functionality inter alia including scheduling of radio resources. The evolved packet core may be responsible for routing calls and data connections to external networks.
The LTE network is a “two node architecture” consisting of serving gateways (SGW) and enhanced base stations, so-called eNode Bs (abbreviated eNB or eNode B). The serving gateways will handle evolved packet core functions, i.e. routing calls and data connections to external networks, and also implement radio access network functions. Thus, the serving gateway may be considered as to combine functions performed by GGSN (Gateway GPRS Support Node) and SGSN (Serving GPRS Support Node) in today's 3G networks and radio access network functions as for example header compression, ciphering/integrity protection. The eNode Bs may handle functions as for example Radio Resource Control (RRC), segmentation/concatenation, scheduling and allocation of resources, multiplexing and physical layer functions.
A mobile communication network is typically modular and it is therefore possible to have several network entities of the same type. The interconnections of network elements are defined by open interfaces. UEs can connect to an eNode B via the air interface or Uu. The eNode Bs have a connection to a serving gateway via the S1 interface. Two eNode Bs are interconnected via the X2 interface.
Both 3GPP and Non-3GPP integration may be handled via the serving gateway's interface to the external packet data networks (e.g. Internet).
QoS Control
Efficient Quality of Service (QoS) support is seen as a basic requirement by operators for LTE. In order to allow best in class user experience, while on the other hand optimizing the network resource utilization, enhanced QoS support should be integral part of the new system.
Aspects of QoS support is currently being under discussion within 3GPP working groups. Essentially, the QoS design for System Architecture Evolution (SAE)/LTE is based on the QoS design of the current UMTS system reflected in 3GPP TR 25.814, “Physical layer aspects for evolved Universal Terrestrial Radio Access (UTRA)”, v.7.1.0 (available at http://www.3gpp.org and incorporated herein by reference). The agreed SAE Bearer Service architecture is depicted in FIG. 2. The definition of a bearer service as given in 3GPP TR 25.814 may still be applicable:
“A bearer service includes all aspects to enable the provision of a contracted QoS. These aspects are among others the control signaling, user plane transport and QoS management functionality”.
In the new SAE/LTE architecture the following new bearers have been defined: the SAE Bearer service between the mobile terminal (User Equipment—UE) and the serving gateway, the SAE Radio Bearer on the radio access network interface between mobile terminal and eNode B as well as the SAE Access Bearer between the eNode B and the serving gateway.
The SAE Bearer Service provides:
QoS-wise aggregation of IP end-to-end-service flows;
IP header compression (and provision of related information to UE);
User Plane (UP) encryption (and provision of related information to UE);
if prioritized treatment of end-to-end-service signaling packets is required an additional SAE Bearer Service can be added to the default IP service;
provision of mapping/multiplexing information to the UE;
provision of accepted QoS information to the UE.
The SAE Radio Bearer Service provides:
transport of the SAE Bearer Service data units between eNode B and UE according to the required QoS;
linking of the SAE Radio Bearer Service to the respective SAE Bearer Service.
The SAE Access Bearer Service provides:
transport of the SAE Bearer Service data units between serving gateway and eNode B according to the required QoS;
provision of aggregate QoS description of the SAE Bearer Service towards the eNode B;
linking of the SAE Access Bearer Service to the respective SAE Bearer Service.
In 3GPP TR 25.814 a one-to-one mapping between an SAE Bearer and an SAE Radio Bearer. Furthermore there is a one-to-one mapping between a radio bearer (RB) and a logical channel. From that definition it follows that a SAE Bearer, i.e. the corresponding SAE Radio Bearer and SAE Access Bearer, is the level of granularity for QoS control in an SAE/LTE access system. Packet flows mapped to the same SAE Bearer receive the same treatment.
For LTE there will be two different SAE bearer types: the default SAE bearer with a default QoS profile, which is configured during initial access and the dedicated SAE bearer (SAE bearers may also be referred to as SAE bearer services) which is established for services requiring a QoS profile which is different from the default one.
The default SAE bearer is an “always on” SAE bearer that can be used immediately after LTE_IDLE to LTE_ACTIVE state transition. It carries all flows which have not been signaled a Traffic Flow Template (TFT). The Traffic Flow Template is used by serving gateway to discriminate between different user payloads. The Traffic Flow Template incorporates packet filters such as QoS. Using the packet filters the serving gateway maps the incoming data into the correct PDP Context (Packet Data Protocol Context). For the default SAE bearer, several service data flows can be multiplexed. Unlike the Default SAE Bearer, the Dedicated SAE Bearers are aimed at supporting identified services in a dedicated manner, typically to provide a guaranteed bit-rate. Dedicated SAE bearers are established by the serving gateway based on the QoS information received in Policy and Charging Control (PCC) rules from evolved packet core when a new service is requested. A dedicated SAE bearer is associated with packet filters where the filters match only certain packets. A default SAE bearer is associated with “match all” packet filters for uplink and downlink. For uplink handling the serving gateway builds the Traffic Flow Template filters for the dedicated SAE bearers. The UE maps service data flows to the correct bearer based on the Traffic Flow Template, which has been signaled during bearer establishment. As for the default SAE Bearer, also for the dedicated SAE Bearer several service data flows can be multiplexed.
The QoS Profile of the SAE bearer is signaled from the serving gateway to the eNode B during the SAE bearer setup procedure. This profile is then used by the eNode B to derive a set of Layer 2 QoS parameters, which will determine the QoS handling on the air interface. The Layer 2 QoS parameters are input the scheduling functionality. The parameters included in the QoS profile signaled on S1 interface from serving gateway to eNode B are currently under discussion. Most likely the following QoS profile parameters are signaled for each SAE bearer: Traffic Handling Priority, Maximum Bit-rate, Guaranteed Bit-rate. In addition, the serving gateway signals to the eNode B the Allocation and Retention Priority for each user during initial access.
Uplink Access Scheme for LTE
For uplink transmission, power-efficient user-terminal transmission is necessary to maximize coverage. Single-carrier transmission combined with FDMA (Frequency Division Multiple Access) with dynamic bandwidth allocation has been chosen as the evolved UTRA uplink transmission scheme. The main reason for the preference for single-carrier transmission is the lower peak-to-average power ratio (PAPR), compared to multi-carrier signals (OFDMA—Orthogonal Frequency Division Multiple Access), and the corresponding improved power-amplifier efficiency and assumed improved coverage (higher data rates for a given terminal peak power). During each time interval, Node B assigns users a unique time/frequency resource for transmitting user data thereby ensuring intra-cell orthogonality. An orthogonal access in the uplink promises increased spectral efficiency by eliminating intra-cell interference. Interference due to multipath propagation is handled at the base station (Node B), aided by insertion of a cyclic prefix in the transmitted signal.
The basic physical resource used for data transmission consists of a frequency resource of size BWgrant grant during one time interval, e.g. a sub-frame of 0.5 ms, onto which coded information bits are mapped. It should be noted that a sub-frame, also referred to as transmission time interval (TTI), is the smallest time interval for user data transmission. It is however possible to assign a frequency resource BWgrant over a longer time period than one TTI to a user by concatenation of sub-frames.
The frequency resource can either be in a localized or distributed spectrum as illustrated in FIG. 3 and FIG. 4. As can be seen from FIG. 3, localized single-carrier is characterized by the transmitted signal having a continuous spectrum that occupies a part of the total available spectrum. Different symbol rates (corresponding to different data rates) of the transmitted signal imply different bandwidths of a localized single-carrier signal.
On the other hand, as shown in FIG. 4, distributed single-carrier is characterized by the transmitted signal having a non-continuous (“comb-shaped”) spectrum that is distributed over system bandwidth. Note that, although the distributed single-carrier signal is distributed over the system bandwidth, the total amount of occupied spectrum is, in essence, the same as that of localized single-carrier. Furthermore, for higher/lower symbol rate, the number of “comb-fingers” is increased/reduced, while the “bandwidth” of each “comb finger” remains the same.
At first glance, the spectrum in FIG. 4 may give the impression of a multi-carrier signal where each comb-finger corresponds to a “sub-carrier”. However, from the time-domain signal-generation of a distributed single-carrier signal, it should be clear that what is being generated is a true single-carrier signal with a corresponding low peak-to-average power ratio. The key difference between a distributed single-carrier signal versus a multi-carrier signal, such as e.g. OFDM (Orthogonal Frequency Division Multiplex), is that, in the former case, each “sub-carrier” or “comb finger” does not carry a single modulation symbol. Instead each “comb-finger” carries information about all modulation symbols. This creates a dependency between the different comb-fingers that leads to the low-PAPR characteristics. It is the same dependency between the “comb fingers” that leads to a need for equalization unless the channel is frequency-non-selective over the entire transmission bandwidth. In contrast, for OFDM equalization is not needed as long as the channel is frequency-non-selective over the sub-carrier bandwidth.
Distributed transmission can provide a larger frequency diversity gain than localized transmission, while localized transmission more easily allows for channel-dependent scheduling. Note that, in many cases the scheduling decision may decide to give the whole bandwidth to a single UE to achieve high data rates.
UL Scheduling Scheme for LTE
The uplink scheme allows for both scheduled access, i.e. controlled by eNode B, and contention-based access.
In case of scheduled access the UE is allocated a certain frequency resource for a certain time (i.e. a time/frequency resource) for uplink data transmission. However, some time/frequency resources can be allocated for contention-based access. Within these time/frequency resources, UEs can transmit without first being scheduled. One scenario where UE is making a contention-based access is for example the random access, i.e. when UE is performing initial access to a cell or for requesting uplink resources.
For the scheduled access Node B scheduler assigns a user a unique frequency/time resource for uplink data transmission. More specifically the scheduler determines
which UE(s) that is (are) allowed to transmit,
which physical channel resources (frequency),
for how long the resources may be used (number of sub-frames)
Transport format (Transport Block Size (TBS) and Modulation Coding Scheme (MCS)) to be used by the mobile terminal for transmission
The allocation information is signaled to the UE via a scheduling grant, sent on the so-called L1/L2 control channel. For simplicity, this downlink channel is referred to the “uplink grant channel” in the following. A scheduling grant message contains at least information which part of the frequency band the UE is allowed to use, the validity period of the grant, and the transport format the UE has to use for the upcoming uplink transmission. The shortest validity period is one sub-frame. Additional information may also be included in the grant message, depending on the selected scheme. Only “per UE” grants are used to grant the right to transmit on the Uplink Shared Channel UL-SCH (i.e. there are no “per UE per RB” grants). Therefore the UE needs to distribute the allocated resources among the radio bearers according to some rules, which will be explained in detail in the next section. Unlike in HSUPA there is no UE based transport format selection. The base station (eNode B) decides the transport format based on some information, e.g. reported scheduling information and QoS information, and UE has to follow the selected transport format. In HSUPA Node B assigns the maximum uplink resource and UE selects accordingly the actual transport format for the data transmissions.
Uplink data transmissions are only allowed to use the time-frequency resources assigned to the UE through the scheduling grant. If the UE does not have a valid grant, it is not allowed to transmit any uplink data. Unlike in HSUPA, where each UE is always allocated a dedicated channel there is only one uplink data channel shared by multiple users (UL SCH) for data transmissions.
To request resources, the UE transmits a resource request message to the Node B. This resources request message could for example contain information on the buffer status, the power status of the UE and some Quality of Services (QoS) related information. This information, which will be referred to as scheduling information, allows Node B to make an appropriate resource allocation. Throughout the document it's assumed that the buffer status is reported for every radio bearer. Of course other configurations for the buffer status reporting are also possible.
Since the scheduling of radio resources is the most important function in a shared channel access network for determining Quality of Service, there are a number of requirements that should be fulfilled by the uplink scheduling scheme for LTE in order to allow for an efficient QoS management (see 3GPP RAN WG#2 Tdoc. R2-062606, “QoS operator requirements/use cases for services sharing the same bearer”, by T-Mobile, NTT DoCoMo, Vodafone, Orange, KPN; available at http://www.3gpp.org/ and incorporated herein by reference):
The UL scheduling scheme for LTE should provide a finer network-based QoS control than what is supported in UMTS Release 6 (HSUPA)
Starvation of low priority services should be avoided
Clear QoS differentiation for radio bearers/services should be supported by the scheduling scheme
The UL reporting should allow fine granular buffer reports (e.g. per radio bearer or per radio bearer group) in order to allow the eNode B scheduler to identify for which Radio Bearer/service data is to be sent.
It should be possible to change the priorities used in the UL scheduling decision of the UE dynamically—based on operator requirements
It should be possible to make clear QoS differentiation between services of different users
It should be possible to provide a minimum bit-rate per radio bearer
As can be seen from above list one essential aspect of the LTE scheduling scheme is to provide mechanisms with which the operator can control the partitioning of its aggregate cell capacity between the radio bearers of the different QoS classes. The QoS class of a radio bearer is identified by the QoS profile of the corresponding SAE bearer signaled from serving gateway to eNode B as described before. An operator can then allocate a certain amount of its aggregate cell capacity to the aggregate traffic associated with radio bearers of a certain QoS class.
The main goal of employing this class-based approach is to be able to differentiate the treatment of packets depending on the QoS class they belong to. For example, as the load in a cell increases, it should be possible for an operator to handle this by throttling traffic belonging to a low-priority QoS class. At this stage, the high-priority traffic can still experience a low-loaded situation, since the aggregate resources allocated to this traffic is sufficient to serve it. This should be possible in both uplink and downlink direction.
One benefit of employing this approach is to give the operator full control of the policies that govern the partitioning of the bandwidth. For example, one operator's policy could be to, even at extremely high loads, avoid starvation of traffic belonging to its lowest priority QoS Class. The avoidance of starvation of low priority traffic is one of the main requirements for the UL scheduling scheme in LTE. In current UMTS Release 6 (HSUPA) scheduling mechanism the absolute prioritization scheme may lead to starvation of low priority applications. E-TFC selection (Enhanced Transport Format Combination selection) is done only in accordance to absolute logical channel priorities, i.e. the transmission of high priority data is maximized, which means that low priority data is possibly starved by high priority data. In order to avoid starvation the Node B scheduler must have means to control from which radio bearers a UE transmits data. This mainly influences the design and use of the scheduling grants transmitted on the L1/L2 control channel in downlink. In the following the details of the UL rate control procedure in LTE is outlined.
Semi-Persistent Scheduling (SPS)
In the downlink and uplink, the scheduling eNode B dynamically allocates resources to user equipments at each transmission time interval via the L1/L2 control channel(s) (PDCCH(s)) where the user equipments are addressed via their specific C-RNTIs. The CRC of a PDCCH is masked with the addressed user equipment's C-RNTI (so-called dynamic PDCCH). Only a user equipment with a matching C-RNTI can decode the PDCCH content correctly, i.e. the CRC check is positive. This kind of PDCCH signaling is also referred to as dynamic (scheduling) grant. A user equipment monitors at each transmission time interval the L1/L2 control channel(s) for a dynamic grant in order to find a possible allocation (downlink and uplink) it is assigned to.
In addition, E-UTRAN can allocate uplink/downlink resources for initial HARQ transmissions persistently. When required, retransmissions are explicitly signaled via the L1/L2 control channel(s). Since retransmissions are scheduled, this kind of operation is referred to as semi-persistent scheduling (SPS), i.e. resources are allocated to the user equipment on a semi-persistent basis (semi-persistent resource allocation). The benefit is that PDCCH resources for initial HARQ transmissions are saved. For details on semi-persistent scheduling, see 3GPP TS 36.300, “Evolved Universal Terrestrial Radio Access (E-UTRA) and Evolved Universal Terrestrial Radio Access Network (E-UTRAN); Overall description; Stage 2 (Release 8)”, version 8.7.0, section 11, January 2009 or 3GPP TS 36.321 “Evolved Universal Terrestrial Radio Access (E-UTRA); Medium Access Control (MAC) protocol specification (Release 8)”, version 8.5.0, section 5.10 March 2009, both available at http://www.3gpp.org and incorporated herein by reference.
One example for a service, which might be scheduled using semi-persistent scheduling is Voice over IP (VoiP). Every 20 ms a VoIP packet is generated at the codec during a talk-spurt. Therefore eNode B could allocated uplink or respectively downlink resource persistently every 20 ms, which could be then used for the transmission of Voice over IP packets. In general, semi-persistent scheduling is beneficial for services with a predictable traffic behavior, i.e. constant bit rate, packet arrival time is periodic.
The user equipment also monitors the PDCCHs in a sub-frame where it has been allocated resources for an initial transmission persistently. A dynamic (scheduling) grant, i.e. PDCCH with a C-RNTI-masked CRC, can override a semi-persistent resource allocation. In case the user equipment finds its C-RNTI on the L1/L2 control channel(s) in the sub-frames where the sub-frame has a semi-persistent resource assigned, this L1/L2 control channel allocation overrides the semi-persistent resource allocation for that transmission time interval and the user equipment does follow the dynamic grant. When sub-frame doesn't find a dynamic grant it will transmit/receive according to the semi-persistent resource allocation.
The configuration of semi-persistent scheduling is done by RRC signaling. For example the periodicity, i.e. PS_PERIOD, of the persistent allocation is signaled within Radio resource Control (RRC) signaling. The activation of a persistent allocation and also the exact timing as well as the physical resources and transport format parameters are sent via PDCCH signaling. Once semi-persistent scheduling is activated, the user equipment follows the semi-persistent resource allocation according to the activation SPS PDCCH every semi-persistent scheduling interval (SPS interval). Essentially the user equipment stores the SPS activation PDCCH content and follows the PDCCH with the signaled periodicity.
In order to distinguish a dynamic PDCCH from a PDCCH, which activates semi-persistent scheduling, i.e. also referred to as SPS activation PDCCH, a separate identity has been introduced in LTE. Basically, the CRC of a SPS activation PDCCH is masked with this additional identity which is in the following referred to as SPS C-RNTI. The size of the SPS C-RNTI is also 16 bits, same as the normal C-RNTI. Furthermore the SPS C-RNTI is also user equipment-specific, i.e. each user equipment configured for semi-persistent scheduling is allocated a unique SPS C-RNTI.
In case a user equipment detects a semi-persistent resource allocation is activated by a corresponding SPS PDCCH, the user equipment will store the PDCCH content (i.e. the semi-persistent resource assignment) and apply it every semi-persistent scheduling interval, i.e. periodicity signaled via RRC. As already mentioned, a dynamic allocation, i.e. signaled on dynamic PDCCH, is only a “one-time allocation”.
Similar to the activation of semi-persistent scheduling, the eNode B also can deactivate semi-persistent scheduling. Semi-persistent scheduling de-allocation is signaled by SPS PDCCH with both the Modulation and coding scheme field and the Resource Block Assignment field bits all set to ‘1’.
For semi-persistent scheduling (SPS) in LTE Release 8, if semi-persistent scheduling is configured and activated, it is assumed that there is only one radio bearer set up which has data suitable for semi-persistent scheduling. For future releases of LTE (e.g. LTE-Advanced) it is assumed that more than one radio bearer suitable for semi-persistent scheduling can be set up, so that semi-persistent scheduling needs to deliver data of more than one radio bearer.
Buffer Status Reporting
The buffer status reporting procedure in LTE is used to provide the eNode B with information about the amount of data available for transmission in the uplink buffers of the user equipments on a per logical channel basis—please note that data of each radio bearer are mapped to are respective logical channel. A so-called Buffer Status Report (BSR) is triggered, if any of the following events happen:
Uplink data, for a logical channel (i.e. of a respective radio bearer) which belongs to a Logical Channel Group (LCG), becomes available for transmission in the RLC (Radio Link Control) or PDCP (Packet Data Convergence Protocol) layer. Furthermore, the data belongs to a logical channel with higher priority than the priorities of the logical channels for which data is already available for transmission. A “Regular BSR” is triggered in this case.
Uplink resources are allocated and the number of padding bits in the transport block (MAC PDU) is equal to or larger than the size of the Buffer Status Report MAC control element. A “Padding BSR” is triggered in this case.
A serving cell change occurs. A “Regular BSR” is triggered in this case.
Furthermore, a (periodic) buffer status report is also triggered by the expiry of the following timers:
when the RETX_BSR_TIMER expires and the UE has data available for transmission a “Regular BSR” is triggered.
when PERIODIC_BSR_TIMER expires, a “Periodic BSR” is triggered.
If a “Regular BSR” or “Padding BSR” was triggered and more than one logical channel group (LCG) has data available for transmission in the referring transmission time interval a so-called “Long BSR” will be sent which is reporting on the buffer status for all four LCGs. In case only one LCG has data available, a so-called “Short BSR” including only the data of this LGC will be sent.
If a “Padding BSR” was triggered, it depends on the amount of padding bits available in the referring transmission time interval, what kind of buffer status report will be sent. If the amount of padding bits is large enough to accommodate a Long BSR, this type of BSR will be sent.
In case that more than one LCG has data in the buffer to report and the amount of padding bits does not allow a Long BSR but there are enough padding bits to send a Short BSR, a so-called “Truncated BSR” is sent. The Truncated BSR has the same format of the Short BSR and reports the LCG that includes the logical channel that has data available for transmission and that has the highest priority.
In case there is only one LCG with data to report and padding bits allow for a Short BSR, a Short BSR is send.
If the buffer status reporting procedure determines that currently a buffer status report has been triggered and the UE has uplink resources allocated for a new transmission in the current transmission time interval a BSR MAC control element is created for inclusion in the current MAC PDU, i.e. the buffer status report is multiplexed with the uplink (user) data. The PERIODIC_BSR_TIMER is restarted every time a BSR is sent, except for situations where a “Truncated BSR” is transmitted.
In case there are no uplink resources allocated for the current transmission time interval and a “Regular BSR” was triggered, a scheduling request (SR) is triggered in order to request uplink resources for transmitting the buffer status report.
In one MAC PDU, there can be at most one MAC BSR control element for sending a buffer status report, even if multiple BSR events occurred. The “Regular BSR” and the “Periodic BSR” have precedence over the “Padding BSR”. The RETX_BSR_TIMER is restarted upon reception of a grant for transmission of the buffer status report.
In case the uplink grant can accommodate all pending data available for transmission but is not sufficient to additionally accommodate the BSR MAC control element, all triggered BSRs are cancelled. Furthermore, all triggered BSRs are cancelled when a buffer status report is included in a MAC PDU for transmission.
Scheduling Requests
The Scheduling Request (SR) is used for requesting resources for a new transmission, e.g. MAC PDU. As indicated above, control information, like a buffer status report, and user data are multiplexed within the MAC PDU. When a scheduling request is triggered, it is considered pending. As long as one scheduling request is pending, the user equipment first checks, if there are uplink resources available on the uplink shared channel (UL-SCH) for a transmission in this transmission time interval. In this case all pending scheduling requests are cancelled.
If there are no uplink resources on the UL-SCH within the next transmission time interval but the user equipment has a valid PUCCH resource for scheduling requests configured for this transmission time interval (and the transmission time interval is not part of a measurement gap), the user equipment transmits a scheduling request on the PUCCH and the SR_COUNTER is incremented.
If SR_COUNTER=SR_TRANS_MAX or if there are no valid PUCCH resources in any transmission time interval, all pending scheduling requests are cancelled and a Random Access procedure is initiated.
Buffer Status Reporting and its Impact on Resource Scheduling
To highlight the problems that may occur in current LTE systems in view of the above outlined buffer status reporting procedure and related scheduling behavior of the eNode Bs, an exemplary scenario is assumed in the following where there exists at least one radio bearer that carries data which is intended to be transmitted on semi-persistently configured uplink resources (semi-persistently scheduled radio bearer). For simplicity of the explanations, it is further assumed that this radio bearer is carrying data of a VoIP (Voice over IP) service. The radio bearer is therefore also referred to in the following as a “VoIP bearer”.
Assuming that the VoIP bearer is the only active bearer and new VoIP data arrives in the UE buffer which was empty previously, the arrival of the new VoIP data will trigger a buffer status report (BSR) for the Logical Channel Group (LCG) the logical channel the VoIP bearer is assigned to, as described above, as can be seen in FIG. 7. The buffer status report triggers a scheduling request including the buffer status report. The scheduling request (SR) is sent to the eNode B on the next uplink control channel (PUCCH) resource. Once the eNode B is informed on the buffer status in the user equipment, the eNode B assigns a dynamic uplink radio resource by means of signaling an uplink grant on the downlink control channel (PDCCH) resources. Four TTIs after reception of the grant, the uplink radio resources are available and the VoIP data can be transmitted to eNode B.
Assuming that there is more than one radio bearer configured and more than one radio bearer assigned to the Logical Channel Group the VoIP bearer is belonging to, the eNode B only knows that the transmitted uplink data originated from the VoIP bearer after reception of the data.
Since IP packets containing VoIP data will undergo header compression (Robust Header Compression (RoHC)—see Bormann et al., IETF RFC 3095, “RObust Header Compression (ROHC): Framework and four profiles: RTP, UDP, ESP, and uncompressed”, available at http://www.ietf.org) in the PDCP layer, the first few VoIP packets of the VoIP bearer may be assumed larger than those following afterwards in steady state operation of the header compression scheme, since the header compression needs to analyze the first packets in order to determine the compression parameters before the compression is activated and before the compression is effective.
The above description is the reason that in the case of VoIP traffic eNode B needs to wait for a few packets until it can determine the compressed VoIP data size in order to allocate the correct size for semi-persistently configured resources.
The above explanations apply e.g. to VoIP data, on the other hand, it is not excluded that radio bearers carrying data suitable to be transmitted on semi-persistently configured resources that show a stable data size from the beginning of the data transmission are not subject to header compression.
Once the semi-persistently configured resources in the uplink are activated and still only one VoIP bearer is actively having data the following scenario can be assumed. VoIP data has a typical periodicity of 20 ms, so the eNode B configures the semi-persistent uplink resources with a periodicity of 20 ms. In order to have low latency it is desirable to have the configured uplink resources available as soon as the VoIP data is arriving in the UE buffer. However, due to the uncertainty of the data arrival of the VoIP packets in the UE buffer, the eNode B cannot exactly determine the TTI where the configured resources should start. Therefore data arrival will be in one of the TTIs in between two configured semi-persistently uplink resources. In a good configuration the arrival of VoIP data packets will be just before the semi-persistently configured uplink resources become available.
Since the VoIP bearer is the only active radio bearer it can be assumed that the configured uplink grant is sufficient to empty the buffer of the UE. This means that all data in the buffer can be transmitted in the semi-persistently configured uplink resource and that the next VoIP data packet will arrive in an empty buffer at UE side. Furthermore, it can be assumed that while the VoIP bearer is the only active bearer, it is not the only radio bearer that is set up by the UE. Since the VoIP data arrives in an empty buffer of the UE, a buffer status report is triggered. This buffer status report becomes available in a TTI where no uplink resources are available. According to the standard LTE specifications this situation triggers a scheduling request to be sent by the UE. The scheduling request is delivered to eNode B at the next available TTI with a configured uplink control channel (PUCCH). The scenario described so far is exemplarily shown in FIG. 8.
Taking the behavior of the current LTE specification the arrival of data which is intended for the semi-persistently scheduled uplink resources is creating an unnecessary buffer status report that is delivered to the eNode B. Since the buffer status report only reports the buffer status per Logical Channel Group, the eNode B might not be aware data of which radio bearer has triggered the buffer status report. Hence, the eNode B cannot be sure that the semi-persistently scheduled resource is sufficient for delivering the data in the UE buffer in the uplink (e.g. the VoIP packet might have arrived after eNode B received the buffer status report). Therefore, the eNode B needs to assign a dynamic uplink resource to UE by means of a dynamic grant in order to assure speedy data delivery. Since dynamically scheduled uplink resources are allocated 4 transmission time intervals after sending the corresponding dynamic grant on the PDCCH, there are two scenarios for the uplink data delivery of the VoIP packet:
The dynamically scheduled resources are available before the semi-persistently scheduled resources: VoIP packet is transmitted according to the dynamic grant, so that the semi-persistently scheduled resources are wasted.
The dynamically allocated resources are available after the semi-persistently scheduled resources: The VoIP packet is transmitted on the semi-persistently scheduled resources, and the dynamically allocated resources are wasted.
In both scenarios the dynamic grant is unnecessary and either the dynamic or the semi-persistently scheduled resources are wasted.
In the following the scenario above is extended to a situation where there are two active VoIP bearers configured at the UE. It is assumed that a first VoIP bearer is already active and semi-persistently scheduled resources are configured for its data—see FIG. 9.
Every time new data from the first VoIP bearer arrives in the UE a buffer status report and a scheduling request is triggered as described above. Once the eNode B receives the buffer status report, it cannot know from which of the two configured VoIP bearers the data reported on in the buffer status report stems from. Hence, the eNode B needs to give a dynamic grant to the UE in order to assure a speedy and correct data delivery.
As can be seen in FIG. 9, if data from the second VoIP bearer arrives in the UE buffer, the buffer is once again empty and a new buffer status report and a scheduling request is triggered. Since the eNode B already received data of the first radio bearer on the dynamically allocated resources, it knows that the data reported in the new buffer status report must be from the second VoIP bearer. If the eNodeB received the buffer status well before the next TTI in which the UE has been assigned semi-persistently allocated resources, the eNode B could override the semi-persistently allocated resources with a dynamic uplink grant that fits the size of the data of the second VoIP bearer, which means that both the data of the first and the second VoIP bearer can be transmitted on the overwritten semi-persistently allocated uplink resources. However, if the buffer status report arrives too late in the eNode B, the eNode B needs to signal an additional dynamic uplink grant that will allocate dynamically allocated resources in a TTI after the TTI where the semi-persistently allocated uplink resources are configured. This can result in unnecessary segmentation and delay to the data of the VoIP bearers.